A fast and high-precision method for detecting unevenness of silicon steel sheets

Abstract

The present invention relates to a silicon steel plate unevenness detection technology, specifically a fast and high-precision silicon steel plate unevenness detection method. The present invention solves the problem that existing silicon steel plate unevenness detection methods cannot take into account both detection accuracy and detection speed. A fast and high-precision silicon steel plate unevenness detection method is implemented by the following steps: Step 1: Use a uniform line light source to illuminate the edge of the silicon steel plate, and use an industrial camera to collect the edge wave image of the silicon steel plate edge; Step 2: Use a seed point growth method that uses brightness gradient as a growth criterion to segment the ROI area of the edge wave image; Step 3: Use a three-scale Gaussian filter to enhance the image of the ROI area; Step 4: Use the maximum inter-class variance method to binarize the ROI area, and then use a contour extraction function to extract the contour line of the ROI area; Step 5: Use the RANSAC algorithm to fit the lower curve of the contour line of the ROI area. The present invention is suitable for silicon steel plate unevenness detection.

Description

Technical Field

[0001] The invention relates to a silicon steel plate unevenness detection technology, in particular to a rapid and high-precision silicon steel plate unevenness detection method. Background Art

[0002] Silicon steel sheets are primarily used to make the cores of various generators, motors, and transformers. They are an important soft magnetic alloy material. During the production process, temperature fluctuations and tension can easily cause edge rippling along the sheets, affecting their quality. Therefore, to ensure the quality of silicon steel sheets, it is necessary to test their roughness (i.e., the ratio of the wave height to the wavelength of the edge rippling).

[0003] Currently, there are two main methods for detecting unevenness in silicon steel sheets: the first is contact detection (e.g., vernier caliper detection), which has the advantage of fast detection speed but suffers from low detection accuracy. The second is non-contact detection (e.g., laser detection), which has the advantage of high detection accuracy but suffers from slow detection speed.

[0004] Based on this, it is necessary to invent a fast and high-precision silicon steel plate unevenness detection method to solve the problem that the existing silicon steel plate unevenness detection method cannot take into account both detection accuracy and detection speed. Summary of the Invention

[0005] In order to solve the problem that the existing silicon steel plate unevenness detection method cannot take into account both detection accuracy and detection speed, the present invention provides a fast and high-precision silicon steel plate unevenness detection method.

[0006] The present invention is achieved by adopting the following technical solutions:

[0007] A fast and high-precision method for detecting unevenness of silicon steel plates is implemented by the following steps:

[0008] Step 1: Use a uniform line light source to illuminate the edge of the silicon steel plate, and use an industrial camera to capture the edge wave image of the silicon steel plate;

[0009] Step 2: Use the seed point growth method with brightness gradient as the growth criterion to segment the ROI area of the edge wave image;

[0010] Step 3: Use three-scale Gaussian filtering to enhance the image of the ROI area; the image enhancement formula is as follows:

[0011] B i =G i *I(i=1,2,3) (1);

[0012] D i =IB i(i=1,2,3) (2);

[0013] D * =(1-0.25*sgn(D1))*D1+0.25*D2+0.25*D3 (3);

[0014] I * =D * +I (4);

[0015] In formulas (1)-(4): B i represents the ROI area after Gaussian filtering at the i-th scale; G i represents the Gaussian kernel of the i-th scale Gaussian filter; I represents the original ROI area; D i Represents the detail information of the ROI area after Gaussian filtering at the i-th scale; D * Represents the overall details of the ROI area; I * Indicates the ROI area after image enhancement;

[0016] Step 4: Use the maximum inter-class variance method to binarize the ROI area, and then use the contour extraction function to extract the contour line of the ROI area;

[0017] Step 5: Use the RANSAC algorithm to fit the lower curve of the contour line of the ROI area into a straight line segment, so that the contour line of the ROI area includes a line segment and an arc; then, on the one hand, the length of the line segment is used as the wavelength of the edge wave, and on the other hand, the distance from each point on the arc to the line segment is calculated, and the maximum value of the distance is used as the wave height of the edge wave; then, the ratio of the wave height of the edge wave to the wavelength is calculated, and the ratio is used as the unevenness of the silicon steel plate, thereby realizing the unevenness detection of the silicon steel plate.

[0018] Compared to existing silicon steel sheet unevenness detection methods, the rapid and high-precision silicon steel sheet unevenness detection method described in the present invention utilizes a completely new principle to achieve unevenness detection of silicon steel sheets. This method offers the following advantages: First, it offers higher detection accuracy than contact-based detection methods. Second, it offers faster detection speed than non-contact detection methods. Therefore, the present invention balances detection accuracy and speed, fully guaranteeing the quality of silicon steel sheets.

[0019] The present invention effectively solves the problem that the existing silicon steel plate unevenness detection method cannot take into account both detection accuracy and detection speed, and is suitable for silicon steel plate unevenness detection. DETAILED DESCRIPTION

[0020] A fast and high-precision method for detecting unevenness of silicon steel plates is implemented by the following steps:

[0021] Step 1: Use a uniform line light source to illuminate the edge of the silicon steel plate, and use an industrial camera to capture the edge wave image of the silicon steel plate;

[0022] Step 2: Use the seed point growth method with brightness gradient as the growth criterion to segment the ROI area of the edge wave image;

[0023] Step 3: Use three-scale Gaussian filtering to enhance the image of the ROI area; the image enhancement formula is as follows:

[0024] B i =G i *I(i=1,2,3) (1);

[0025] D i =IB i (i=1,2,3) (2);

[0026] D * =(1-0.25*sgn(D1))*D1+0.25*D2+0.25*D3 (3);

[0027] I * =D * +I (4);

[0028] In formulas (1)-(4): B i represents the ROI area after Gaussian filtering at the i-th scale; G i represents the Gaussian kernel of the i-th scale Gaussian filter; I represents the original ROI area; D i Represents the detail information of the ROI area after Gaussian filtering at the i-th scale; D * Represents the overall details of the ROI area; I * Indicates the ROI area after image enhancement;

[0029] Step 4: Use the maximum inter-class variance method to binarize the ROI area, and then use the contour extraction function to extract the contour line of the ROI area;

[0030] Step 5: Use the RANSAC algorithm to fit the lower curve of the contour line of the ROI area into a straight line segment, so that the contour line of the ROI area includes a line segment and an arc; then, on the one hand, the length of the line segment is used as the wavelength of the edge wave, and on the other hand, the distance from each point on the arc to the line segment is calculated, and the maximum value of the distance is used as the wave height of the edge wave; then, the ratio of the wave height of the edge wave to the wavelength is calculated, and the ratio is used as the unevenness of the silicon steel plate, thereby realizing the unevenness detection of the silicon steel plate.

[0031] The uniform line light source adopts a red high-brightness uniform line light source; the industrial camera adopts a wide-field-of-view, high-resolution industrial camera.

[0032] The contour extraction function adopts the findContours function of OpenCV.

[0033] Although specific embodiments of the present invention have been described above, those skilled in the art will appreciate that these are merely illustrative and that the scope of the present invention is defined by the appended claims. Those skilled in the art may make various changes or modifications to these embodiments without departing from the principles and essence of the present invention, and such changes and modifications are intended to fall within the scope of the present invention.

Claims

1. A fast and high-precision method for detecting unevenness of silicon steel sheets, characterized by: This method is implemented using the following steps: Step 1: Use a uniform line light source to illuminate the edge of the silicon steel plate, and use an industrial camera to capture the edge wave image of the silicon steel plate; Step 2: Use the seed point growth method with brightness gradient as the growth criterion to segment the ROI area of the edge wave image; Step 3: Use three-scale Gaussian filtering to enhance the image of the ROI area; the image enhancement formula is as follows: B i =G i *I(i=1,2,3) (1); D i =I-B i (i=1,2,3) (2); D * =(1-0.25*sgn(D1))*D1+0.25*D2+0.25*D3 (3); I * =D * +I (4); In formulas (1)-(4): B i represents the ROI area after Gaussian filtering at the i-th scale; G i represents the Gaussian kernel of the i-th scale Gaussian filter; I represents the original ROI area; D i Represents the detail information of the ROI area after Gaussian filtering at the i-th scale; D * Represents the overall details of the ROI area; I * Indicates the ROI area after image enhancement; Step 4: Use the maximum inter-class variance method to binarize the ROI area, and then use the contour extraction function to extract the contour line of the ROI area; Step 5: Use the RANSAC algorithm to fit the lower curve of the contour line of the ROI area into a straight line segment, so that the contour line of the ROI area includes a line segment and an arc; then, on the one hand, the length of the line segment is used as the wavelength of the edge wave, and on the other hand, the distance from each point on the arc to the line segment is calculated, and the maximum value of the distance is used as the wave height of the edge wave; then, the ratio of the wave height of the edge wave to the wavelength is calculated, and the ratio is used as the unevenness of the silicon steel plate, thereby realizing the unevenness detection of the silicon steel plate.

2. A rapid and high-precision silicon steel sheet unevenness detection method according to claim 1, characterized in that: The uniform line light source adopts a red high-brightness uniform line light source; the industrial camera adopts a wide-field-of-view, high-resolution industrial camera.

3. A rapid and high-precision silicon steel sheet unevenness detection method according to claim 1, characterized in that: The contour extraction function adopts the findContours function of OpenCV.